cancel
Showing results for 
Search instead for 
Did you mean: 

BIA Accelerator and Multi cube Queries

Former Member
0 Kudos

Hi guys,

Can we apply BIA indexing for Multi cube queries?..

I think multi cube is no way realted with BIA indexes, as no data is stored in them.

but i would like the expert opinion that whether BIA indexes improve the performance of the queries built on Multicubes and by indexing the cubes used in that multi cubes.

Thanks in advance.

Loki.

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Build the indexes on the basic cube that is part of the multicube andte BIA will read.

Ravi Thothadri

Former Member
0 Kudos

Hi ,

I know query performace is incresed by BIA , but how come the load performace is increased using BIA , i guess to eliminate aggregates roll up, is there anything i am missing here, can someone clarify.

Thanks

Vj

Former Member
0 Kudos

In BIA you "aggregate" the whole InfoCube, where is with a classic aggregate you aggregate to a specific subset of data. The load performance is increased due to the technology used by mutiple blade servers and TREX coding. If you are loading mutiple aggregates for InfoCubes, BIA is therefore faster overall since your basically building aggregates against all subsets of data. (which is really the beauty of BIA) The real advantage is you also have many more CPU/Memory dedicated to saving this data to the BIA NFS. Utlimately this really helps the maintenance as opposed to aggregates since you don't have to roll-up several (and some cases hundreds or thousands) or aggregates per InfoCube.

Hope that helps.

Former Member
0 Kudos

Hi Mike,

Can you more specific about "<b>BIA you "aggregate" the whole InfoCube</b>",

My question is that can't we apply BIA indexing only to certain requets as we do for aggregates?If so can you mention the steps as well.

also can we maintain both Aggregates & BIA indexes for a cube simultaneously?

Thank you in advance.

Regds/Loki.

Former Member
0 Kudos

Hi Loki,

You cannot do selective requests aggregattion with BIA, it has to be full data in the infocube, you can still use aggregates with BIA but the query will first access Cache , then BIA and then it goes to Aggregates. so if you have BIA on the Cube then its of no use to maintain Aggregates.

Please correct me if i am wrong.

Thanks

Vj

Former Member
0 Kudos

Ok, so when you create an aggregate you are basically selecting a fraction of an InfoCube based on a certain set of variables (this is the importat part) and that aggregate is more or less a smaller InfoCube and a subset of that original InfoCube.

In BIA, in a way, we aggregate every variable possibility.

You can continue to maintain aggregates (roll-ups) in case BIA fails for some odd reason but if BIA is connected it will use them first.

Answers (3)

Answers (3)

Former Member
0 Kudos

MultiProviders: any indexed InfoCube as part of the MultiProvider will use SAP NetWeaver BI Accelerator. Note that overall query performance is defined by the slowest InfoCube in a MultiProvider environment .

Hope it Helps

Chetan

@CP..

Former Member
0 Kudos
Former Member
0 Kudos

Hello Loki,

changerun and rollup into the BIA index is faster than into aggregates therefore the loading performance is also improved. If the cube is stored in the BIA and your loading schedule is really tight you can also consider to keep the database indexes on the cube dropped. Avoiding the index rebuild after the load will save you save you some more minutes. You can do this because the BI system will read all data from the BIA and not from the database when a query is executed. But I would only use such tricks if you have to squeeze out every second.

Best regards,

Ralf

Former Member
0 Kudos

Hi,

What if the Cube is not for querying but for loading the ODS from Cube, even in this case BIA helps ? , and how does the loading into ODS effect, and is there any place where we can see the ODS is reading from BIA not from Cube, if there is anythign like happens ?

Thanks

Vj.

Former Member
0 Kudos

ODS are not currently optimized for BIA.

In general I think there is some confusion. We are talking about two things here. BIA is a means of indexing data currently in an InfoCube. It is a replacement for the use of aggregates, not for the use of data-staging containers (ODS, InfoCube, etc). It has its advantages over aggregates since it is much faster at doing the 'roll-up' process, not the data staging process.